Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
2000-04-06
2004-11-23
Kelley, Chris (Department: 2713)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
C375S240040, C375S240060
Reexamination Certificate
active
06823009
ABSTRACT:
FIELD OF THE INVENTION
The subject of the present invention is a method for evaluating the degradation of a video image introduced by a coding and/or storage and/or digital transmission system, particularly a system employing digital video signal transmission and/or storage and/or coding with a low throughput.
BACKGROUND OF THE INVENTION
The degradation is generally due to the errors introduced by the throughput-reduction algorithms and/or by defective transmission links or even to defects in the coders and decoders.
The digitizing of video signals has opened the possibility of being able to copy, store or transmit the video information while maintaining constant quality.
However, the large amount of information transported by the video images in practice requires the use of digital compression methods in order to reduce the binary throughput.
One compression method which is very widespread in the video field is described in the ISO/CEI 13918 MPEG2 standard. This algorithm is said to be of the “lossy” type, since the image reproduced after decoding is not identical to the original. In order to maintain quality which is acceptable to the final viewer, the throughput-reduction algorithms take account of the perceptual properties of the human system of vision. Despite that, the constraints imposed, of throughput or of available bandwidth for the transmission, as well as the content of the video signal, imply the appearance of characteristic degradation in the signal after decoding. This degradation, introduced by the system, for example MPEG2, has a direct influence on the perceived quality of the final image.
Automatic evaluation of the quality of audio-visual signals has a wide range of applications in the digital television system: production, distribution and evaluation of the performance of the systems.
Furthermore, the existing devices were derived for laboratory tests and are not suitable for remote monitoring of the distribution networks.
Sequences of MPEG-coded images transmitted at low data rate for broadcasting digital television or for other multimedia applications will exhibit a certain number of defects or deformations with respect to the original sequence. In fact, a non-exhaustive list of visible degradations could be drawn up. Those of them which are most perceptible are granular errors, deformation of contours, losses of information, the “exotic” contours, block effects, etc. However, small-scale transmission errors may be conveyed by more or less localized effects on the image. In the event of a significant disturbance, they may be expressed as difficulties of access to the information, for example breaks in service or freezing of images for a longer or shorter time depending on the disturbance. The scope of the errors depends on the level of relevance and of the structure of the data which they affect: synchronization words, movement vectors, images coded with or without prediction or basic images for the predictions. In addition to breaks or freezing of images, the degradation observed is expressed as blocks or macroblocks which are erroneous or incorrectly positioned. This has the effect of propagating the degradation over the entire video sequence up to the image coded without prediction, hence coded independently of the others.
One evaluation method has already been proposed by the NTIA (National Telecommunications and Information Administration) in the article by A. A. WEBSTER and Colleagues, entitled “An objective video quality assessment system based on human perception” and published in June 1993 in the magazine SPIE, vol. 13, p. 15-26.
This method employs an analysis of the degraded images and of the original images, after they have been filtered by two vertical and horizontal SOBEL operators (3×3 matrices). The filtered image is obtained by convolution, by making the SOBEL matrices slide horizontally and vertically, and the results obtained represent the vertical and horizontal gradients of the image. In other words, the filtered image highlights the vertical and horizontal contours contained in the initial (unfiltered) image.
A measurement based on this information makes it possible to highlight the change in content between the input to the video system and its output.
The method proposed by the NTIA employs two parameters:
on the one hand, the spatial information SI which represents the standard deviation measured on the pixels of the filtered image via the SOBEL operator. It is a question here of determining the standard deviation at the level of the contours of the filtered image, considering that the contours are important for viewing and that they are affected by the various processing operations of the throughput-reduction digital systems;
on the other hand, the time-based information TI which represents the standard deviation of the difference image between two successive images, this standard deviation being calculated on the basis of the differences between the values of the same pixels of two successive frames. The parameter TI may reveal a jerky movement due to a defect in the coder.
The method proposed by the NTIA employs a comparative calculation of the parameters SI and TI over the digital video signal, between an input image and an output image of a system.
This method exhibits a certain number of drawbacks.
The SOBEL filter conserves only certain frequencies of the image which make it possible to take account of the loss of contours, which means that the loss of definition which could be taken into account is that which is situated in the range of frequencies conserved. In other words, the loss of definition can be only partially taken into account.
Moreover, the parameter SI takes account of defects which tend to compensate out. In fact, the loss of information from the image tends to make the parameter SI diminish, whereas the false contours and the block effects tend, in contrast, to make it increase, which means that the parameter SI is meaningful only if one or other of the phenomena is dominant.
Finally, the method of calculating the SI and TI parameters, by using a standard deviation calculated in overall terms on the entire image, drastically reduces the impact on these parameters of localized degradations.
SUMMARY OF THE INVENTION
The subject of the present invention is a method which allows the abovementioned defects to be at least partially remedied.
The method according to the present invention makes use of a block transform, for example the discrete cosine transform used particularly in the MPEG standard, in order to highlight the characteristic signatures of the defects identified. This original approach makes it possible not only to make fine measurements on the errors introduced, but also makes it possible to take account of the initial content of the video signal and of the algorithms employed in MPEG.
The block transformations of an image (Fourier transform, discrete cosine transform DCT, etc.) are obtained via the operation:
[Fn,m]=[T].[fn,m].[T]
t
in which f(x,y) designates the image block to be transformed and T(x,y) the matrix of the transformation. Another block transform is produced on the basis of the wavelet-transform of the image, by reorganizing the wavelet coefficients in such a way as to obtain transformed blocks having the desired size, and in particular of the same size as the blocks obtained by the abovementioned methods. Such a method of reorganization is set out in the article by R. de QUEIROZ and Colleagues, entitled “Wavelet transforms in a JPEG-like Image Coder”, published in April '1997 in the magazine IEEE Transactions On Circuits and Systems for Video Technology, vol. 7 no. 2 p. 419-424.
The basic idea of the invention is, in particular, to carry out calculations on the blocks according to which the transmitted image was coded, in such a way as to generate a meaningful parameter exempt from the block effect.
The invention thus relates to a method of evaluating the degradation of a video image coded by blocks of image points or pixels
Baina Jamal
Bretillon Pierre
Kelley Chris
Michael & Best & Friedrich LLP
Teledifusion de France
Wong Allen
LandOfFree
Method for evaluating the degradation of a video image... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for evaluating the degradation of a video image..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for evaluating the degradation of a video image... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3282956